Solid versus Shell Elements 3

[IJR]

I am not very familiar with solid elements in structural analysis FE software.

I will appreciate any basic information and practical application of SOLID element modelling, with emphasis on software use.

respects
IJR

 

[brad]

I don't mean to offend, but as worded this is a pretty basic question which requires a very long answer if one is to be thorough.

The only succinct answer which I can give is that solid elements are used widely in all general FEA solver codes. They are used for a variety of things, and are extremely common.

I would recommend that you refer to an introductory FEA text. A few introductory texts which I have found useful:
Chandrupatla and Belegundu "Introduction to Finite Elements in Engineering"
Cook, "Finite Element Modeling for Stress Analysis".

If you have more specifically-worded questions, I may be able to readily answer them, but this question is very wide in its scope.

Brad

 

[Qshake]

IJR, Brad's right on the money with this one. I've have also found those texts useful and, in fact, have a copy of the Chandrupatla text.

If you've something more specific...Brad is the guy to look to for to-the-point answers.

 

[IJR]

Thanks bradh and Qshake for the fast responses

Bradh: IJR believes in democracy and is rarely offended. Feel free to spit in.

I have a book on finite element analysis and I think I have a little bit of feel for what the stuff is.

Is Chandrupatla's book good enough to be put into use fast?
The book I have has very few numerical examples.

Thanks
IJR

 

[brad]

IJR,
I've got both books at work, but unfortunately I'm typing this from home on a Friday evening. My recollection is that the Cook book has more practical examples than Chandrupatla. Chandrupatla is loaded more with the basics of formulation (written at a senior level/non-mechanics grad level). I don't recall that either gives a lot of detail to some of the more important pitfalls of 'real' usage (hourglassing, shear locking, et al.).

What is the basis for your question? It is just curiousity, or do have a specific concern about solid elements? I'm happy to spout about aspects of solid elements, but if you let me go about it unchecked I may keep you reading for awhile (I've never been accused of being quiet . . .).

Brad

 

[IJR]

Thanks Bradh for keeping the faith

I do have concern for solid elements and I am familiar with finite element basics. My current problem is that I have never used solid elements before. So you see it is more a question of how to apply it than what it is.

If you wish you could go on give some ancyclopaedic info, like where you find using them very handy, some watch outs etc. The rest I can handle.

Current project: A 200year old stadium built from stones and is to be rehabilitated and capacity to be increased. So a friend of mine has to analyse the stony structure.

Thanks once more and again
IJR

 

[anandaganesh]

You can refer to the book Finite element Method by Zeinkwicz which is equally good.

 

[brad]

IJR,
I just got done going over both Chandrupatla/Belegundu and Cook, and neither really has examples along the lines of real application details. I have yet to find a book that does a very good job of these things. Perhaps one day I'll write one and make a lot of money . . .

So, lacking a good source, let me spout a bit (now that I better understand your issues).

I'll talk about solid elements by first discussing 'structural' elements (beams and shells). Beams are appropriate in the context of beam theory, and shells in the context of shell theory. However, many structures satisfy neither beam theory nor shell theory. In these cases, the solid elements are often most suited.

Solid elements are basically the most 'general purpose' of any of the elements. If one could model something with beam elements or shell elements, then one could almost surely model it with solid elements. The reverse, however, often does not hold true.

In your case of the stadium, there are probably many features which do not readily lend themselves to either of the specific theories, which is why you are asking the question.

I am actually a big fan of solid elements. With a good pre-processor, just about anything can be meshed with solids. However, there are some things to be wary of:

1) 1st order solid elements do not behave well in bending. Fully-integrated 1st order solids have overly-stiff behavior (known as 'shear locking'). One must be wary of this behavior. Many codes have addressed this problem with element formulations. Just be aware of this (please look around on the FEA strings for prior discussions on this--I have posted on this subject before).

1b) 1st order reduced-integration elements actually shear too much (called 'hourglassing'), so this can be a problem. Again, look around on this site for some other postings about this.

2) Solid elements do not react moments at their nodes, so any moments must be addressed either via force-couples or by attaching structural elements or constraints which then 'smear' the moment into multiple nodes of the solid elements.

3) One should try to avoid having only one element through the thickness. I in fact try to have at least 4 elements of thickness through any features (a general rule of thumb). Any less than this can have difficulty accurately describing the deformation pattern (resulting in errant results).

Hopefully this is along the lines of what you are looking for. Fire off any more questions you may have.

Regards,
Brad

 

[IJR]

Bradh

Your post is exactly what I was looking for. Thanx once more and again. If you are serious on writing a book, then go ahead do it- I have a feeling you can do it.

respects
IJR

 

[brad]

Glad to be of service.

One more thing that I didn't mention:
My discussion on '1st order elements' was meant to imply first order hexahedral elements ('cubes'). In case you are not aware:

NEVER NEVER NEVER use a mesh significantly comprised of 1st order tetrahedral elements.

Just to drive it home-NEVER!

All of the warnings about triangular shell elements hold, plus many more. 1st order tetrahedrals are bad news. They are grossly over-stiff, and will result in shockingly bad answers (it's the nature of the beast, nothing you can do about it). The only situation in which I would ever use them is in a 1st order hexahedral/pentahedral mesh that needs one in a transition zone (but then, if it's a complex mesh I'll use 2nd order tets). In nine years of FEA, I could probably count on both hands the total number of 1st tets that I've built. Treat them like radioactive sludge.

Brad